JP3951732B2 - Image processing method, image processing apparatus, image processing program, and recording medium - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an image processing method, an image processing apparatus, an image processing program, and a recording medium for extracting a desired region from an image.
[0002]
[Prior art]
In general, as the (extracted) technique such taking out areas such as object or a person from the image, for example, 63-115279 Patent and Publication Sho include techniques disclosed in JP-A-10 -63855 . Among them, the technique described in the former publication calculates the contour of the object included in the image, calculates each area of the region delimited by the contour, and processes the region having the maximum area value as the background. The technique described in the latter publication divides an image into areas having pixels of similar colors, and allows the user to specify an area to be extracted from among these areas, and This technique also extracts a color region that approximates the region.
[0003]
[Problems to be solved by the invention]
However, in the technique of the former publication, it is not possible to extract an image (for example, a close-up image) in which the area having the maximum area is not the background, and in the technique of the latter publication, the boundary of the extracted area is not There was a drawback of being ambiguous. Thus, it can be said that there is still room for improvement as a technique for extracting a region from an image.
The present invention has been made in view of such circumstances, and an object of the present invention is to extract a desired region from an image with a clear boundary and less influence of details. An image processing method, an image processing apparatus, an image processing program, and a recording medium are provided.
[0004]
[Means for Solving the Problems]
In order to achieve the above object, an image processing method according to the present invention includes a detection step of luminance difference of pixels adjacent in the input image to detect a line connecting between the pixel exceeding the threshold value as a contour, the There is a difference in luminance between tiles that are tiled with a plurality of pixels in the input image that have a predetermined brightness so as not to include a contour line and between tiles that face each other across the contour line. When a determination process for determining whether or not the value is within the threshold value and the result of the determination is affirmative, tiles facing each other across the outline and pixels sandwiched between the tiles are defined as tiles. collectively, and the combined tile, while associated with the same area with each other, and when the determination result is negative, a different region of the facing tiles to each other across the contour Seki Together give, among among the tiles, from pixels adjacent to one of the tiles to the pixel that are separated by the contour associated with the which is one tile in the same region, among each other the tile, adjacent the other tiles And an associating process for associating the pixels from the pixel to the pixels delimited by the contour line with the same area as the other tile .
According to the present invention, the outline of the image is detected, along with the pixel to be free of the contour line is tiled, the tile between delimited by contours, the pixel value is within the threshold If there are, they are associated as the same region. Here, among the pixels sandwiched between the tiles, the peripheral pixels adjacent to the tiles to the pixels immediately before the contour line are associated with the tiles, so that the boundary portion between the regions does not become rough. Furthermore, in the portion other than the contour lines, a plurality of pixels are tiled gathered, it reduces the effect of details such as dust.
[0005]
In the present invention, in the associating process, it is preferable that the luminance of a pixel sandwiched between tiles for which the determination result is negative is estimated and changed from the luminance of adjacent tiles . Accordingly, each of the pixels is close to the color of the adjacent tile, so that the vicinity of the contour can be expressed more finely.
[0006]
In the present invention, even if the result of the determination is negative, if the user designates, the process of re-associating tiles facing each other across the contour line with each other as the same region is performed after the association process. It is preferable to have. For example, the boundary between the bottom surface and the side surface of the cone in the image is detected as a contour line , but if the degree of reflection is different, it may not be associated as the same region. Even in such a case, the bottom surface and the side surface can be extracted as the same cone.
[0007]
In the present invention, the luminance of the tile is preferably a value obtained by averaging the luminances of a plurality of pixels collected as a tile. In addition, in order to reflect the user's intention in the processed image, the user has a mode in which the user presets the luminance difference between the tiles at least before the determination step, and the pixel size to be combined in one tile is determined by the user. Is preferable to have a preset process at least before the tiling process.
[0008]
In addition to the above-described image processing method, the present invention is an image processing apparatus, an image processing program for causing a computer to function as the means as the image processing apparatus, and a recording medium on which the program is recorded In any case, a desired region can be extracted from an image with clear boundaries and less influence of details.
[0009]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments of the present invention will be described with reference to the drawings.
[0010]
<Overall configuration>
FIG. 1 is a diagram illustrating an overall configuration of an image processing apparatus according to an embodiment. As shown in this figure, the image processing apparatus 100 includes a personal computer (PC) 102, a display 104, a keyboard 106, a mouse 108, a printer 110, and the like.
Among these, the PC 102 is an execution subject of an image processing program to be described later, and is connected to the network 50 by a built-in modem or a LAN card, and can exchange data with the server 60 and the like. The display 104 functions as a display screen output unit in the PC 102, and the keyboard 106 and the mouse 108 function as input units for the PC 102, respectively. The printer 110 functions as an output unit that prints an image processed by the PC 102 on paper or the like.
[0011]
FIG. 2 is a block diagram showing a functional configuration constructed when the PC 102 executes the image processing program. The image processing program may be downloaded from the server 60 via the network 50 and installed on the PC 102, or may be installed on the PC 102 via a medium such as a magnetic disk, an optical disk, or a magneto-optical disk. good.
[0012]
In FIG. 2, an input unit 150 corresponds to the keyboard 106 and mouse 108 in FIG. 1, and is used for designating an image to be processed, image processing operation / setting, and the like. The input image storage unit 152 stores image data that can be processed. Here, the processing target includes an image read by a flatbed scanner (not shown), an image downloaded from the server 60, an image stored in various media such as a magneto-optical disk, and the like.
[0013]
The image extraction unit 154 is constructed by executing an image processing program, and decomposes and extracts an image to be processed into regions. The details of the image area decomposition / extraction will be described later.
The extracted image storage unit 156 stores the image regions extracted by the image extraction unit 154, the association of these regions, and the like. The input image storage unit 152 and the extracted image storage unit 156 correspond to a storage unit (for example, a hard disk) built in the PC 102, but such a function is provided to other devices (for example, the server 60) via the network 50. It is good also as a structure (network drive) which makes it burden.
The output unit 158 corresponds to the display 104 or the printer 110, and outputs an image area stored in the extracted image storage unit 156 or an image area specified by the user among image areas associated with each other.
[0014]
<Contents of image processing>
Next, processing contents of the image processing program executed in the PC 102 will be described. FIG. 3 is a flowchart showing details of the image processing program. First, immediately after the user starts the image processing program, for example, the PC 102 displays a list of images that can be processed on the display 104 and a screen that prompts the user to select an image to be processed. When an image to be actually processed is selected by the user, in step S11, the PC 102 reads image data corresponding to the image.
If the image data to be loaded is a format described in PDL (Page Description Language) or a vector format, the PC 102 converts the image data into a raster (bitmap) format indicated by a set of pixels. .
[0015]
In step S <b> 12, the PC 102 (image extraction unit 154) detects the outline of an object (including a person) included in the image while retaining the raster format image data as it is. This contour detection is executed by the following algorithm, for example. That is, when the difference between the pixel value of the target pixel and the pixel value of the adjacent pixel exceeds a preset threshold value, it is determined that the object contour is between the target pixel and the adjacent pixel. . Here, the pixel value includes not only a coordinate value that defines the color of the pixel in various color spaces but also a value indicating luminance.
[0016]
Here, when the contour is detected on the raster image as shown in FIG. 4A, the contour line is shifted by half a pixel in the lower right direction as shown in FIG. 4B. Assuming a line (white thick broken line), pixels located on the line are displayed as, for example, black as contour pixels.
4 (a) and 4 (b), only the arrangement of the pixels is shown for the sake of explanation, and information on the colors inherent to the pixels is omitted except for the black color described here ( The same applies to FIGS. 5 and 6). Further, since the raster format image data is held as it is separately from the display of the contour, just because the contour pixel is blackened, the information about the original color is not lost.
[0017]
Next, in step S13, the PC 102 executes tiling. In this tiling, first, as shown in FIG. 5 (a), the retained raster image is considered as one block every 3 pixels × 3 pixels, and secondly, as shown in FIG. 5 (b). As described above, when a block includes a contour pixel, the block is left as it is. When the block does not include a contour pixel, the nine pixels constituting the block are grouped into one tile. It is.
In the present embodiment, as a method of combining nine pixels into one tile, in this embodiment, a value obtained by averaging the pixel values of the pixels for the nine pixels constituting the tile is obtained, and the average value is calculated as the nine pixels. It is supposed to reset to a new pixel value. As a result, the nine pixels have the same color, which is substantially the same as the process of replacing the nine pixels with one new pixel.
[0018]
When extracting a fine region, it is desirable to reduce the size of one tile, and when extracting a large region that wraps a fine portion, it is desirable to increase the size of one tile. . Therefore, after detecting the outline and before tiling, the PC 102 causes the display 104 to display a screen that prompts the user to select an area to be extracted, analyzes the characteristics of the specified area, and determines the tile size. The optimum value may be automatically set, or may be set by the user at least before tiling, and the tiling may be executed according to the size of the set value.
[0019]
After the tiling, in step S14, the PC 102 compares the pixel values of the adjacent tiles and the tiles facing each other across the outline pixel, and determines whether or not to associate the tiles with each other. A process for determining (a tile combining process) is executed.
Here, the PC 102 compares the pixel values of adjacent tiles, and if the difference is within a threshold value, the PC 102 associates the tiles with the same region, When the difference exceeds the threshold, the tiles are associated with different areas.
Further, the PC 102 compares the pixel values of tiles facing each other with the outline pixel therebetween, and if the difference is within a threshold value, the PC 102 associates the tiles with the same region, The pixels sandwiched between the tiles are also tiled and associated with the same area as that associated with the tiles. On the other hand, if the pixel value difference exceeds the threshold value, the tiles are different areas. And the pixels sandwiched between the tiles are left without being tiled.
[0020]
The association between the same region and the different region is performed by, for example, assigning an identifier for individually identifying the region to each of the tile or the pixel (not tiled). . That is, the same identifier is assigned to tiles or pixels that display the same area, and different identifiers are assigned to tiles or pixels that display different areas.
Here, the pixel values of adjacent tiles and the tiles facing each other across the contour pixel are compared to determine whether or not the tiles are associated with each other. Therefore, without comparing pixel values (or considering the comparison result between pixel values), the region closed by the detected contour is set as the same region, but is separated by a certain contour. These two regions may be associated as different regions.
[0021]
Next, in step S15, the PC 102 executes interpolation / association processing for pixels that have not been tiled. That is, in this process, pixels that are not tiled because they include pixels that display contours at the time of blocking, and the pixel values of the tiles that face each other across the pixel have a threshold value. Interpolation processing for a pixel (non-tile pixel) that has been left because it has exceeded, and processing for associating with a region.
Specifically, the PC 102 firstly calculates a contour line from the arrangement of pixels for displaying a contour, and secondly determines which region the non-tile pixel is associated with by the contour line, and associates it. Interpolate from the pixel values of tiles in the specified area.
[0022]
For example, when tiled as shown in FIG. 6A, a partially enlarged view of a partial region A is as shown in FIG. 6B. In FIG. 6B, the contour pixel is indicated by hatching, but the contour line B is indicated by a broken line when calculated backward from the contour pixel arrangement. That is, since the pixel located on the line obtained by shifting the contour line by a half pixel in the lower right direction is the contour pixel, the line connecting the centers of the contour pixels is shifted by the half pixel in the upper left direction. Things become the contour line B. Since the tile has already been associated with any region, here, the peripheral pixels adjacent to a certain tile to the pixel immediately before the contour line B are associated with the same region as the tile.
[0023]
Therefore, in FIG. 6B, when the tile located outside the contour line B is associated with the region c and the tile located inside the contour line B is associated with the region d, Pixels located outside the contour line B are associated with the region c, and pixels located inside the contour line B are associated with the region d. Further, the pixel value of the associated pixel is obtained by interpolation in consideration of the pixel value of the adjacent tile. For this reason, since the finer pixel value of the non-tile pixel is calculated from the tiled portion, the fineness of the boundary portion is ensured. In the interpolation, the pixel value of the already interpolated pixel may be taken into consideration.
[0024]
Next, in step S16, the PC 102 temporarily stores the image data in a state where the tile and the pixel are associated with each other.
In step S <b> 17, the PC 102 determines whether or not it is specified that at least two or more areas are to be combined with the same area. The reason for determining here is, for example, that when a solid exists in the input image, each side of the solid is detected as an outline in step S12, so that each surface of the solid is a pixel value even if it is the same solid. Are different from each other, it is sometimes associated with different regions. Even in such a case, the user does not extract each surface of the solid individually, This is because there is a case where it is desired to extract as a region, and it is necessary to leave a room where the user's intention is reflected to some extent.
[0025]
If the determination result is affirmative, the PC 102 associates the designated areas with the same area in step S18. Specifically, the PC 102 changes the identifiers of tiles and pixels constituting another designated region to the identifiers of tiles and pixels constituting one designated region. As a result, the designated areas are re-associated with each other as being the same area.
Thereafter, the PC 102 returns the processing procedure to step S16 again, and overwrites the image data with the changed association state.
On the other hand, when there is no area combination designation by the user in step S17, the image processing is ended.
[0026]
<Specific examples of image processing>
Next, a specific example of image processing by this image processing program will be described. First, as shown in FIG. 7, when an image including a triangular pyramid P1, a cylinder P2, and a rectangular parallelepiped P3 is read and processed (step S11), as shown in FIG. Is detected (step S12).
Subsequently, for the image to be processed, pixels are blocked and blocks that do not include contour pixels are tiled, while blocks that do not include contour pixels are left alone (step 13). Pixel values such as each other are compared, and those within the threshold value are associated as the same region (step S14). As a result, as shown in FIG. 9, each surface of the triangular pyramid P1, each surface of the cylinder P2, each surface of the rectangular parallelepiped P3, and their backgrounds are associated with each other as different regions. In FIG. 9, the small rectangle is a tile, and the pixel is much smaller than that, and is not shown in the figure. Here, the outline is displayed, but the outline may not be displayed.
[0027]
When the association is made for the tile, the non-tile pixel is interpolated / associated (step S15). The details are as described above. The pixels from the peripheral pixel of the block to the front of the contour line are associated with the region associated with the tile adjacent to the peripheral pixel, and the pixel value is interpolated. As a result, as shown in FIG. 10, the non-tile pixel is interpolated by the pixel value of the adjacent tile. After the image data is stored in this state (step S16), it is determined whether or not the user has designated to combine the regions (step S17).
[0028]
In this stored image data, each face of the triangular pyramid P1, each face of the cylinder P2, each face of the rectangular parallelepiped P3, and their backgrounds are associated with different regions. In some cases, instead of extracting each surface of P1 separately, it may be desirable to extract the triangular pyramid P1 formed by these surfaces as a unit. In such a case, the user instructs to join the faces of the triangular pyramid P1. By this instruction, the tiles and pixels constituting each face of the triangular pyramid P1 are changed to have the same identifier (step S18), and the image data in the changed state is overwritten (step S16).
Furthermore, when the user wants to extract the rectangular parallelepiped P3 as one body and instructs to combine the faces of the rectangular parallelepiped P3, the identifiers of tiles and pixels constituting each face of the rectangular parallelepiped P3 are changed to be the same. (Step S18), the image data in the changed state is overwritten (step S16).
[0029]
In this way, each time a region combination is designated, the processing procedure circulates through steps S16, S17, and S18. On the other hand, when region combination is not designated (or when it is confirmed that no region designation is performed thereafter), This image processing ends. For this reason, the image data is stored in the state of being overwritten last.
When the user wants to extract a desired area from the image indicated by the stored image data, an identifier associated with the area is given from all the tiles and pixels constituting the image. Only tiles and pixels need to be extracted.
Therefore, after this image processing is completed, another application (for example, a printing program) is started, the processed image data that has been associated is opened, and a screen that prompts the user to specify a desired area is displayed. In addition, by extracting the tiles and pixels to which the identifiers associated with the designated area are extracted, only the area can be displayed on the display 104 or printed on the printer 110.
For example, when the user wants to extract only the triangular guess P1 and the rectangular parallelepiped P3 in the above-described image example, tiles and pixels to which identifiers corresponding to the faces of the triangular guess P1 are assigned, and the faces of the rectangular parallelepiped P3. By executing the process of extracting each of the tile and the pixel assigned the identifier corresponding to, it is possible to obtain an image in which only a desired region is extracted as shown in FIG.
[0030]
According to the image processing apparatus according to the present embodiment, the pixels are tiled so as not to include the pixels detected as the contour of the object included in the image. This tiling makes the effects of dust and noise less noticeable. Furthermore, the pixels near the contour that are not tiled are related to the region associated with the tile adjacent to the peripheral pixel from the peripheral pixel to the pixel in front of the contour line. Are interpolated, so that even if the contour is not clearly apparent, it is possible to extract a region finely.
[0031]
In the above-described embodiment, the pixels positioned on the line obtained by shifting the detected contour line by a half pixel in the lower right direction are displayed as the contour. However, the inner side or the outer side of the region delimited by the detected contour line The pixels from the peripheral pixel to the front of the contour line are associated with the region associated with the tile adjacent to the peripheral pixel. With respect to the relationship with the pixels to be displayed, if there is any agreement, it is possible to follow this agreement. For example, a pixel located on the outermost side or the innermost side in the region divided by the contour line may be used as the contour pixel. In the first place, since it is ambiguous which region the contour pixel belongs to, it may be possible to select an option of discarding the contour pixel and associating / interpolating only other non-tile pixels with the region.
For interpolation of non-tile pixels, the pixel value of the pixel before processing may be left as it is, and the pixel value is corrected in consideration of the pixel value of the pixel before processing together with the pixel value of the tile. It is also good.
[0032]
【The invention's effect】
As described above, according to the present invention, it is possible to extract a desired region from an image with clear boundaries and less influence of details.
[Brief description of the drawings]
FIG. 1 is a diagram illustrating a configuration of an image processing apparatus according to an embodiment of the present invention.
FIG. 2 is a block diagram showing a functional configuration of the image processing apparatus.
FIG. 3 is a flowchart showing an image processing operation in the image processing apparatus.
4A and 4B are diagrams for explaining contour detection in the image processing apparatus, respectively. FIG.
FIGS. 5A and 5B are diagrams for explaining tiling in the image processing apparatus. FIG.
FIGS. 6A and 6B are diagrams for explaining interpolation of non-tile pixels in the image processing apparatus, respectively.
FIG. 7 is a diagram for explaining an image by the image processing.
FIG. 8 is a diagram for explaining an image by the image processing.
FIG. 9 is a diagram for explaining an image by the image processing.
FIG. 10 is a diagram for explaining an image by the image processing.
FIG. 11 is a diagram for explaining an image by the image processing.
[Explanation of symbols]
100: Image processing apparatus 102: PC
104 ... Display 110 ... Printer 150 ... Input unit 152 ... Input image storage unit 154 ... Image extraction unit 156 ... Extracted image storage unit 158 ... Output unit

Claims (9)

A detection process for detecting a line connecting pixels whose luminance difference between adjacent pixels exceeds a threshold in the input image as a contour line ;
A tiling process to make a tile having a predetermined brightness by collecting a plurality of pixels in the input image so as not to include the contour line ;
A discriminating process for discriminating whether or not a difference in luminance between tiles facing each other across the outline is within a threshold value;
When the judgment result is affirmative, the tiles facing each other across the contour, collectively pixels sandwiched by each among the tiles to the tile, and the combined tile is identical areas from each other While associating with
When the determination result is negative, the tiles facing each other across the contour line are associated with different regions, and the tiles are separated from the pixels adjacent to one tile by the contour line. Are associated with the same area as the one tile, and among the tiles, pixels adjacent to the other tile to pixels separated by the contour line are the same area as the other tile. And an associating process for associating with the image processing method. 2. The image processing according to claim 1, wherein in the associating process, the luminance of a pixel sandwiched between tiles for which the determination result is negative is estimated and changed from the luminance of adjacent tiles. Method. Even if the result of the determination is negative, if specified by the user, a process of re-associating tiles facing each other across the contour line with each other as the same area is included after the association process. The image processing method according to claim 1, wherein: The image processing method according to claim 1, wherein the luminance of the tile is a value obtained by averaging the luminances of a plurality of pixels collected as a tile. The image processing method according to claim 1, further comprising: a step in which the user presets the threshold value at least before the determination step. The image processing method according to claim 1, further comprising: a step in which a user presets pixel sizes to be combined into one tile at least before the tiling step. Detecting means for detecting, as an outline, a line connecting pixels whose luminance difference between adjacent pixels exceeds a threshold in the input image;
Tiling means that tiles the pixels in the input image into a plurality of pixels with a predetermined brightness so as not to include the contour line ; and
A discriminating means for discriminating whether or not a difference in luminance between tiles facing each other across the outline is within a threshold value;
When the judgment result is affirmative, the tiles facing each other across the contour, collectively pixels sandwiched by each among the tiles to the tile, and the combined tile is identical areas from each other While associating with
When the determination result is negative, the tiles facing each other across the contour line are associated with different regions, and the tiles are separated from the pixels adjacent to one tile by the contour line. Are associated with the same area as the one tile, and among the tiles, pixels adjacent to the other tile to pixels separated by the contour line are the same area as the other tile. And an associating means for associating with the image processing apparatus. Computer
Detecting means for detecting, as an outline, a line connecting pixels whose luminance difference between adjacent pixels exceeds a threshold in the input image;
Tiling means that tiles the pixels in the input image into a plurality of pixels with a predetermined brightness so as not to include the contour line ; and
A discriminating means for discriminating whether or not a difference in luminance between tiles facing each other across the contour line is within a threshold value; and
When the judgment result is affirmative, the tiles facing each other across the contour, collectively pixels sandwiched by each among the tiles to the tile, and the combined tile is identical areas from each other While associating with
When the determination result is negative, the tiles facing each other across the contour line are associated with different regions, and the tiles are separated from the pixels adjacent to one tile by the contour line. Are associated with the same area as the one tile, and among the tiles, pixels adjacent to the other tile to pixels separated by the contour line are the same area as the other tile. the image processing program for functioning as an association means for associating with. Computer
Detecting means for detecting, as an outline, a line connecting pixels whose luminance difference between adjacent pixels exceeds a threshold in the input image;
Tiling means that tiles the pixels in the input image into a plurality of pixels with a predetermined brightness so as not to include the contour line ; and
A discriminating means for discriminating whether or not a difference in luminance between tiles facing each other across the contour line is within a threshold value; and
When the judgment result is affirmative, the tiles facing each other across the contour, collectively pixels sandwiched by each among the tiles to the tile, and the combined tile is identical areas from each other While associating with
When the determination result is negative, the tiles facing each other across the contour line are associated with different regions, and the tiles are separated from the pixels adjacent to one tile by the contour line. Are associated with the same area as the one tile, and among the tiles, pixels adjacent to the other tile to pixels separated by the contour line are the same area as the other tile. The computer-readable recording medium which recorded the image processing program for functioning as a correlation means linked | related with.

Images